Well-conditioned robust observer for detection of faulty systems components

Abstract

A new concept of fault detection and isolation using robust observation for systems with random noises is presented. The method selects the parameters from components that may fault during the process and constructs well conditioned robust observers, considering sensors faults. To isolate component failures via robust observation, a bank of detection observers is constructed, where each observer is only sensitive to one specified component failure while robust to all other component failures.